Did you know that the VMware Tanzu Observability by Wavefront engineering team not only listens to input from customers, but acts on that input to improve the overall Tanzu Observability experience? In fact, as a direct result of customer surveys, the Wavefront engineering team recently completed 30 days of improvement focused on query quality. I will quickly run through those improvements in this post.
The toil of maintaining systems
Prior to joining VMware, I was responsible for a team that monitored more than 700 services and thousands of hosts and associated networking devices. The platform that monitored the hosts and provided the network operations console was an off-the-shelf package from a prominent Silicon Valley-based company. It ran entirely on-premises, which required significant effort to manage the environment in the form of servers/hosts, databases, network configuration, and associated licensing. To take advantage of new features or performance improvements, we had to spend time and budget planning and executing upgrades, which distracted us from our core mission.
With VMware Tanzu Observability by Wavefront, you get access to new features and performance improvement with zero distractions. Importantly, Wavefront recently added the ability to ingest Prometheus data and write PromQL in it directly. Other features, such as improved access control and enhancements to distributed tracing, are all in our pipeline; you will get them without any action on your part.
Highlights of 30 days of improvement
Here is a rundown of the improvements the Wavefront engineering team made to Tanzu Observability:
-
Scale – One of Wavefront's largest customers recently scaled its data 10x without issue or any effort on its part, demonstrating Wavefront's commitment to query quality regardless of scale.
-
Data shape/cardinality – Most agree it's a good idea to avoid needlessly high cardinality data sets. The Wavefront team understands, however, that sometimes they’re necessary, and so has patented technology that autonomously inspects the data and shards as needed to keep query response time low. This functionality was further enhanced during the 30 days of improvement to speed up query performance.
-
Query planning – Wavefront now has the ability to convert negated tags, which filter results, to positive tags, which provide hints to the query planner and open additional paths. For example, a negative tag such as "…not DEPLOYMENTGROUP=prod" filters query results after the query has executed. In our example, "…not DEPLOYMENTGROUP=prod" might convert to "DEPLOYMENTGROUP=dev or DEPLOYMENTGROUP=test". By providing more concise input to the query planner, Wavefront can improve query performance.
-
Tag equivalents – Wavefront now also holds patents on matching data similarity rather than how it is named. For example, Wavefront can expect a single host to emit both the packets that are sent and received on a network interface.
-
Joins – Single-side joins improvement now takes place even faster after we memoized the point stream.
-
TopK – The performance of strategy optimization for TopK calculations to ensure fast query response is now even better.
Why you should care
Tanzu Observability ensures you don’t have to spend time maintaining complex environments so you can focus on running your business. Nor do you have to incur costs or expend effort to access new features or take advantage of performance improvements. The Tanzu Observability team is committed to continuously improving the performance and query quality, to asking interesting questions of your data, and to helping you provide the best service to your customers.
Visit https://tanzu.vmware.com/observability to sign up for a free trial and start reducing your toil today!
